Mine prospecting prediction method based on machine learning

Abstract

The invention belongs to the technical field of geological exploration and data mining, and discloses a prospecting prediction method based on machine learning. Comprising the following steps: preprocessing multi-source geological data to obtain a standardized geological data set; the multi-source geological data comprises geophysical data, geochemical data and remote sensing data; performing feature extraction on the standardized geological data set based on a preset feature extraction algorithm to obtain a multi-dimensional feature set; constructing a geological spatial distribution feature map, performing spatial enhancement on the multi-dimensional feature set based on the geological spatial distribution feature map, training the spatial enhancement feature set based on a preset machine learning model, and predicting geological data of the target area by using a prospecting prediction model to obtain a prospecting prediction result; the prospecting prediction result comprises a mineralization probability distribution diagram and an uncertainty distribution diagram; and performing prospecting target optimization based on the prospecting prediction result to obtain an optimized prospecting target, thereby realizing prospecting prediction based on machine learning.

Description

Technical Field

[0001] This invention relates to the field of geological exploration and data mining technology, and more specifically, to a mineral exploration prediction method based on machine learning. Background Technology

[0002] With the continuous growth of global demand for mineral resources, traditional mineral exploration methods face increasing challenges in terms of efficiency and accuracy. Traditional mineral exploration mainly relies on the experience and judgment of geological experts, combined with field geological surveys, geophysical exploration, geochemical analysis, and remote sensing technology, to identify potential mineralized areas through manual analysis and comprehensive evaluation. However, this method often suffers from the following problems when faced with complex geological environments and large-scale data: First, data processing efficiency is low, as traditional methods struggle to quickly integrate and analyze multi-source heterogeneous geological data; second, prediction accuracy is limited, as manual analysis is easily constrained by subjective experience, making it difficult to fully uncover deep patterns and regularities in the data; and third, costs are high, as field exploration and data acquisition require significant investment of manpower, resources, and time, especially in remote or geologically challenging areas.

[0003] In recent years, the rapid development of machine learning technology has provided new solutions for mineral exploration prediction. Machine learning can automatically extract features and build predictive models by learning from large amounts of historical data, thereby improving the efficiency and accuracy of mineral exploration. Currently, some studies have attempted to apply machine learning to the field of mineral exploration prediction. For example, support vector machines (SVM) are used to classify geochemical data to identify potential mineralization anomaly areas; or random forest (RF) models are used to perform regression analysis on geophysical data to predict the depth and size of ore bodies. However, these methods still have many shortcomings in practical applications: First, the data integration capability is insufficient. Existing methods are mostly focused on the analysis of single-type data, making it difficult to effectively integrate the heterogeneous features of multi-source geological data (such as geophysical, geochemical, and remote sensing data); second, the model generalization ability is limited. Existing models are often trained for specific mineral types or specific geological environments, making it difficult to adapt to the prediction needs of different mineral types and geological backgrounds; third, there is a lack of quantification of uncertainty. Existing methods often cannot provide reliable confidence assessments in the prediction results, leading to high decision-making risks.

[0004] Furthermore, with the development of remote sensing and high-resolution geophysical exploration technologies, the scale and complexity of geological data are constantly increasing. How to efficiently process this high-dimensional, multimodal data and fully consider the spatial distribution characteristics and uncertainties of geological data in predictions is a key problem that urgently needs to be solved in the field of mineral exploration prediction. Traditional machine learning methods are prone to the "curse of dimensionality" when processing high-dimensional data and have poor robustness to data noise and missing values. At the same time, existing methods often lack effective utilization of prior geological knowledge during model training and optimization, leading to prediction results that may deviate from actual geological patterns.

[0005] Based on the above problems, there is an urgent need to propose a mineral exploration prediction method based on machine learning, which can efficiently integrate multi-source geological data, construct a prediction model with strong generalization ability and quantifiable uncertainty, thereby improving the accuracy and efficiency of mineral exploration prediction and providing a scientific basis for mineral resource exploration. Summary of the Invention

[0006] To overcome the aforementioned shortcomings of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a mineral exploration prediction method based on machine learning, comprising:

[0007] Acquire multi-source geological data and preprocess the multi-source geological data to obtain a standardized geological dataset; the multi-source geological data includes geophysical data, geochemical data and remote sensing data;

[0008] Based on a preset feature extraction algorithm, features are extracted from a standardized geological dataset to obtain a multidimensional feature set.

[0009] A geological spatial distribution feature map is constructed, and the multidimensional feature set is spatially enhanced based on the geological spatial distribution feature map to obtain a spatially enhanced feature set;

[0010] A mineral exploration prediction model is obtained by training the spatial enhancement feature set based on a pre-set machine learning model.

[0011] The geological data of the target area are predicted using a mineral exploration prediction model to obtain mineral exploration prediction results; the mineral exploration prediction results include mineralization probability distribution maps and uncertainty distribution maps;

[0012] Based on the mineral exploration prediction results, the mineral exploration target area is optimized to obtain the optimized mineral exploration target area.

[0013] Preferably, the preprocessing of the multi-source geological data to obtain a standardized geological dataset includes:

[0014] Data cleaning is performed on multi-source geological data to obtain cleaned geological data; data cleaning includes missing value imputation, outlier removal, and noise filtering.

[0015] The cleaned geological data is converted to a uniform format to obtain geological data.

[0016] Standardized geological datasets are obtained by standardizing geological data in a uniform format; the standardization process includes normalization and coordinate system alignment.

[0017] Preferably, the step of extracting features from the standardized geological dataset based on a preset feature extraction algorithm to obtain a multidimensional feature set includes:

[0018] The dimensionality reduction of the standardized geological dataset is performed based on a pre-defined principal component analysis algorithm to obtain a dimensionality-reduced feature set.

[0019] Based on a pre-defined correlation analysis algorithm, the dimensionality-reduced feature set is filtered to obtain the filtered feature set.

[0020] Based on a preset feature combination algorithm, the selected feature set is combined to obtain a multidimensional feature set.

[0021] Preferably, the step of constructing a geological spatial distribution feature map and spatially enhancing the multidimensional feature set based on the geological spatial distribution feature map to obtain a spatially enhanced feature set includes:

[0022] Geological spatial distribution feature maps are constructed based on standardized geological datasets; these maps include spatial autocorrelation feature maps and spatial heterogeneity feature maps.

[0023] Based on a preset spatial convolution algorithm, spatial enhancement is performed on the multidimensional feature set and the geological spatial distribution feature map to obtain a spatially enhanced feature set.

[0024] Preferably, the step of training the spatial enhancement feature set based on a preset machine learning model to obtain a mineral exploration prediction model includes:

[0025] The initial prediction model is obtained by initially training the spatial augmentation feature set based on the preset deep learning model.

[0026] The hyperparameters of the initial prediction model are optimized based on the preset Bayesian optimization algorithm to obtain the optimized prediction model;

[0027] Uncertainty assessment of the optimized prediction model is performed based on a pre-defined uncertainty quantification algorithm to obtain a mineral exploration prediction model; the uncertainty quantification algorithm includes the Monte Carlo dropout method.

[0028] Preferably, the step of optimizing the prospecting target area based on the prospecting prediction results to obtain the optimized prospecting target area includes:

[0029] Preliminary prospecting target areas were determined based on mineralization probability distribution maps;

[0030] Risk assessment of the preliminary prospecting target area was conducted based on the uncertainty distribution map, and the risk assessment results were obtained.

[0031] The preliminary prospecting target area is optimized based on the risk assessment results and the preset optimization algorithm to obtain the optimized prospecting target area; the preset optimization algorithm includes the genetic algorithm.

[0032] An electronic device, comprising:

[0033] Memory, used to store computer programs;

[0034] A processor is used to execute computer programs.

[0035] A computer-readable storage medium for storing computer programs.

[0036] The technical effects and advantages of the mineral exploration prediction method based on machine learning in this invention are as follows:

[0037] By preprocessing and extracting features from multi-source geological data, heterogeneous data can be efficiently integrated, overcoming the limitations of traditional methods and existing machine learning methods in analyzing single data types, and significantly improving data utilization and prediction accuracy. Through deep learning models combined with hyperparameter optimization and spatial augmentation techniques, a mineral exploration prediction model with strong generalization capabilities is constructed, capable of adapting to the prediction needs of different mineral types and geological backgrounds, thus solving the problem of existing methods' strong dependence on specific scenarios.

[0038] The Monte Carlo dropout method was introduced to assess the uncertainty of the prediction results, generating an uncertainty distribution map. This provides a reliable risk assessment basis for mineral exploration decisions and significantly reduces exploration risks. A genetic algorithm was used to optimize the preliminary mineral exploration target area, comprehensively considering mineralization probability and uncertainty, achieving a balance between exploration costs and prediction effectiveness, and greatly improving the scientific rigor and practicality of the mineral exploration target area.

[0039] By using techniques such as feature dimensionality reduction and spatial convolution, the computational complexity of data processing and model training is reduced. Attached Figure Description

[0040] Figure 1 This is a schematic diagram illustrating the steps of a machine learning-based mineral exploration prediction method according to the present invention.

[0041] Figure 2 This is a structural diagram of an electronic device according to the present invention. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] Example 1

[0044] See also Figure 1 and Figure 2 As shown, this embodiment provides a mineral exploration prediction method, apparatus, equipment, and medium based on machine learning, which addresses the shortcomings of traditional mineral exploration methods, such as low efficiency, limited accuracy, and high cost, as well as the deficiencies of existing machine learning methods in multi-source data integration, model generalization ability, and uncertainty quantification. Specifically, it may include the following steps:

[0045] Step S101: Acquire multi-source geological data and preprocess the multi-source geological data to obtain a standardized geological dataset; the multi-source geological data includes geophysical data, geochemical data and remote sensing data.

[0046] Multi-source geological data refers to geological information collected from different sources and modes, including but not limited to geophysical data (such as gravity data, magnetic data, and seismic data), geochemical data (such as elemental content of soil samples and rock sample analysis data), and remote sensing data (such as hyperspectral and multispectral imagery). This data can be acquired through field surveys, laboratory analysis, and remote sensing satellites. To ensure data quality and consistency, multi-source geological data preprocessing is necessary. Specifically, the preprocessing process includes the following sub-steps:

[0047] S1011, Data cleaning is performed on multi-source geological data to obtain cleaned geological data. Data cleaning includes missing value imputation, outlier removal, and noise filtering. For missing values, K-nearest neighbor (KNN) based interpolation methods can be used for imputation; for outliers, statistical Z-score methods can be used for detection and removal; for noise, wavelet transform based filtering methods can be used for smoothing.

[0048] S1012, the cleaned geological data is converted to a unified format to obtain geological data. Since the formats of multi-source geological data may differ (such as raster format for geophysical data, tabular format for geochemical data, and image format for remote sensing data), it is necessary to convert these data into a unified format, for example, converting all data into a raster data format based on a Geographic Information System (GIS) for subsequent processing.

[0049] S1013 standardizes the uniform-format geological data to obtain a standardized geological dataset. Standardization includes normalization and coordinate system alignment. Normalization uses the min-max normalization method to scale the data to the [0,1] interval. Coordinate system alignment uses a projection transformation based on the WGS84 coordinate system to ensure that all data are aligned in the same spatial reference system.

[0050] This results in a high-quality, standardized geological dataset, laying the foundation for subsequent feature extraction and model training.

[0051] Step S102: Based on the preset feature extraction algorithm, feature extraction is performed on the standardized geological dataset to obtain a multidimensional feature set.

[0052] Because standardized geological datasets contain high-dimensional, multimodal data, directly using them for model training can lead to high computational complexity and model overfitting. Therefore, feature extraction algorithms are needed to extract key features from the data, reduce dimensionality, and improve feature expressiveness. The feature extraction process includes the following sub-steps:

[0053] S1021, based on a preset principal component analysis algorithm, performs dimensionality reduction processing on the standardized geological dataset to obtain a dimensionality-reduced feature set. The principal component analysis (PCA) algorithm projects high-dimensional data into a low-dimensional space, retaining principal components with a cumulative variance contribution rate of 95%, thereby reducing data dimensionality and redundant information.

[0054] S1022, Based on a preset correlation analysis algorithm, feature selection is performed on the dimensionality-reduced feature set to obtain a selected feature set. Pearson correlation coefficient analysis is used to calculate the correlation between each feature and known mineralized areas, selecting features with an absolute correlation coefficient greater than 0.5 and removing features with low correlation.

[0055] S1023, Based on a preset feature combination algorithm, the selected feature set is combined to obtain a multidimensional feature set. A polynomial combination method is used to generate interaction terms between features (such as the product of feature A and feature B), thereby enhancing the expressive power of the features and ultimately obtaining the multidimensional feature set.

[0056] The above methods can extract highly discriminative multidimensional feature sets from high-dimensional geological data, providing high-quality input data for subsequent spatial augmentation and model training.

[0057] Step S103: Construct a geological spatial distribution feature map, and perform spatial enhancement on the multidimensional feature set based on the geological spatial distribution feature map to obtain a spatially enhanced feature set.

[0058] The spatial distribution characteristics of geological data (such as spatial autocorrelation and spatial heterogeneity) are crucial for mineral exploration prediction. To fully utilize this spatial information, it is necessary to construct a geological spatial distribution feature map and spatially enhance the multidimensional feature set. The spatial enhancement process includes the following sub-steps:

[0059] S1031, a geological spatial distribution feature map is constructed based on a standardized geological dataset; the geological spatial distribution feature map includes a spatial autocorrelation feature map and a spatial heterogeneity feature map. The spatial autocorrelation feature map uses Moran's I method to quantify the spatial autocorrelation of each feature; the spatial heterogeneity feature map uses the local entropy method to quantify the spatial heterogeneous distribution of each feature.

[0060] S1032, Spatial enhancement is performed on the multidimensional feature set and the geological spatial distribution feature map based on a preset spatial convolution algorithm to obtain a spatially enhanced feature set. A spatial convolution algorithm based on a convolutional neural network (CNN) is used to convolve the multidimensional feature set with the spatial autocorrelation feature map and the spatial heterogeneity feature map, thereby incorporating spatial distribution information into the feature set to obtain the spatially enhanced feature set.

[0061] The above methods can effectively enhance the ability of feature sets to express geological spatial patterns, thereby improving the accuracy of mineral exploration prediction.

[0062] Step S104: Train the spatial enhancement feature set based on the preset machine learning model to obtain the mineral exploration prediction model.

[0063] To construct a mineral exploration prediction model with high generalization ability and uncertainty quantification capability, a deep learning model combined with hyperparameter optimization and uncertainty quantification methods is used for training. The training process includes the following sub-steps:

[0064] S1041. An initial prediction model is obtained by initially training the spatial enhancement feature set based on a pre-defined deep learning model. A hybrid model based on convolutional neural networks (CNN) and long short-term memory networks (LSTM) is adopted, where CNN is used to extract spatial features and LSTM is used to capture time series features (such as the sampling time series of geochemical data). During training, known mineralized area data are used as positive samples and non-mineralized area data are used as negative samples. The cross-entropy loss function is used, and the Adam optimizer is used.

[0065] S1042, the hyperparameters of the initial prediction model are optimized based on a preset Bayesian optimization algorithm to obtain an optimized prediction model. The Bayesian optimization algorithm is used to optimize the model's hyperparameters (such as learning rate, convolutional kernel size, and number of hidden layer neurons), with the optimization objective being to maximize the F1 score on the validation set.

[0066] S1043. Based on a pre-defined uncertainty quantification algorithm, the uncertainty of the optimized prediction model is evaluated to obtain a mineral exploration prediction model. The uncertainty quantification algorithm includes the Monte Carlo dropout method. In the model inference stage, the Monte Carlo dropout method is used to perform multiple forward propagations on the model to calculate the mean and variance of the prediction results, thereby quantifying the uncertainty of the prediction and finally obtaining a mineral exploration prediction model with uncertainty assessment capabilities.

[0067] This allows us to obtain a mineral exploration prediction model with strong generalization ability and quantifiable uncertainty, providing a reliable basis for subsequent predictions.

[0068] Step S105: Use the mineral exploration prediction model to predict the geological data of the target area to obtain the mineral exploration prediction results; the mineral exploration prediction results include a mineralization probability distribution map and an uncertainty distribution map.

[0069] Geological data of the target area is input into a mineral exploration prediction model for prediction, yielding the prediction results. The prediction results consist of two parts: first, a mineralization probability distribution map, representing the mineralization probability at each location within the target area, with probability values ​​ranging from [0,1]; and second, an uncertainty distribution map, representing the degree of uncertainty of the prediction results at each location, using standard deviation as the uncertainty measure. The prediction results can be visualized using a Geographic Information System (GIS) for analysis and decision-making by geological experts.

[0070] Step S106: Optimize the prospecting target area based on the prospecting prediction results to obtain the optimized prospecting target area.

[0071] To further optimize the prospecting target area based on the prediction results, it is necessary to comprehensively consider mineralization probability and uncertainty. The optimization process includes the following sub-steps:

[0072] S1061, Determine preliminary mineral exploration target areas based on mineralization probability distribution maps. Set a mineralization probability threshold (e.g., 0.8), and classify areas with probability values ​​greater than this threshold as preliminary mineral exploration target areas.

[0073] S1062, a risk assessment is conducted on the preliminary prospecting target area based on the uncertainty distribution map, and the risk assessment results are obtained. The mean uncertainty at each location within the preliminary prospecting target area is calculated as the risk assessment result; the lower the mean uncertainty, the more reliable the prediction result and the lower the risk.

[0074] S1063, based on the risk assessment results and a preset optimization algorithm, the preliminary prospecting target area is optimized to obtain the optimized prospecting target area; the preset optimization algorithm includes a genetic algorithm. The genetic algorithm is used to optimize the preliminary prospecting target area, with the optimization objective being to maximize the total mineralization probability while minimizing the total uncertainty under a given exploration cost constraint. Individuals in the genetic algorithm represent the spatial extent of the target area, and the fitness function is the weighted sum of the total mineralization probability and the total uncertainty; the weights can be adjusted according to actual needs.

[0075] Furthermore, embodiments of the present invention also provide an electronic device, Figure 2 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of the invention. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the machine learning-based mineral prospecting prediction method disclosed in any of the foregoing embodiments. Furthermore, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0076] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this invention, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0077] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0078] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the machine learning-based mineral prospecting prediction method executed by the electronic device 20 as disclosed in any of the foregoing embodiments, the computer program 222 may further include computer programs capable of performing other specific tasks.

[0079] Furthermore, the present invention also provides a computer-readable storage medium for storing a computer program; wherein, when executed by a processor, the computer program implements the aforementioned disclosed machine learning-based mineral prospecting prediction method. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0080] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for users of ordinary technical skills, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A mineral exploration prediction method based on machine learning, characterized in that, include: Acquire multi-source geological data and preprocess the multi-source geological data to obtain a standardized geological dataset; the multi-source geological data includes geophysical data, geochemical data and remote sensing data; Based on a preset feature extraction algorithm, features are extracted from a standardized geological dataset to obtain a multidimensional feature set. A geological spatial distribution feature map is constructed, and the multidimensional feature set is spatially enhanced based on the geological spatial distribution feature map to obtain a spatially enhanced feature set; A mineral exploration prediction model is obtained by training the spatial enhancement feature set based on a pre-set machine learning model. The geological data of the target area are predicted using a mineral exploration prediction model to obtain mineral exploration prediction results; the mineral exploration prediction results include mineralization probability distribution maps and uncertainty distribution maps; Based on the mineral exploration prediction results, the mineral exploration target area is optimized to obtain the optimized mineral exploration target area.

2. The mineral exploration prediction method based on machine learning according to claim 1, characterized in that, The preprocessing of the multi-source geological data to obtain a standardized geological dataset includes: Data cleaning is performed on multi-source geological data to obtain cleaned geological data; data cleaning includes missing value imputation, outlier removal, and noise filtering. The cleaned geological data is converted to a uniform format to obtain geological data. Standardized geological datasets are obtained by standardizing geological data in a uniform format; the standardization process includes normalization and coordinate system alignment.

3. The mineral exploration prediction method based on machine learning according to claim 2, characterized in that, The standardized geological dataset is subjected to feature extraction based on a preset feature extraction algorithm to obtain a multidimensional feature set, including: The dimensionality reduction of the standardized geological dataset is performed based on a pre-defined principal component analysis algorithm to obtain a dimensionality-reduced feature set. Based on a pre-defined correlation analysis algorithm, the dimensionality-reduced feature set is filtered to obtain the filtered feature set. Based on a preset feature combination algorithm, the selected feature set is combined to obtain a multidimensional feature set.

4. The mineral exploration prediction method based on machine learning according to claim 3, characterized in that, The process involves constructing a geological spatial distribution feature map and then spatially enhancing the multidimensional feature set based on the geological spatial distribution feature map to obtain a spatially enhanced feature set, including: Geological spatial distribution feature maps are constructed based on standardized geological datasets; these maps include spatial autocorrelation feature maps and spatial heterogeneity feature maps. Based on a preset spatial convolution algorithm, spatial enhancement is performed on the multidimensional feature set and the geological spatial distribution feature map to obtain a spatially enhanced feature set.

5. The mineral exploration prediction method based on machine learning according to claim 4, characterized in that, The process of training the spatial enhancement feature set based on a preset machine learning model to obtain a mineral exploration prediction model includes: The initial prediction model is obtained by initially training the spatial augmentation feature set based on the preset deep learning model. The hyperparameters of the initial prediction model are optimized based on the preset Bayesian optimization algorithm to obtain the optimized prediction model; Uncertainty assessment of the optimized prediction model is performed based on a pre-defined uncertainty quantification algorithm to obtain a mineral exploration prediction model; the uncertainty quantification algorithm includes the Monte Carlo dropout method.

6. The mineral exploration prediction method based on machine learning according to claim 5, characterized in that, The optimization of the prospecting target area based on the prospecting prediction results to obtain the optimized prospecting target area includes: Preliminary prospecting target areas were determined based on mineralization probability distribution maps; Risk assessment of the preliminary prospecting target area was conducted based on the uncertainty distribution map, and the risk assessment results were obtained. The preliminary prospecting target area is optimized based on the risk assessment results and the preset optimization algorithm to obtain the optimized prospecting target area; the preset optimization algorithm includes the genetic algorithm.

7. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing a computer program to implement the machine learning-based mineral exploration prediction method as described in any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that, Used to store computer programs; wherein, when executed by a processor, the computer programs implement the machine learning-based mineral exploration prediction method as described in any one of claims 1 to 6.

Images